Merge d87fa5f905 into 6d32c897ac

to get other links if we start getting blocked

.gitignore

@@ -3,7 +3,7 @@ node_modules/
 *.db-shm
 *.db-wal
 target/
-.env.*
+*.local
 
 
 .DS_Store

db-datos/.env.development

@@ -0,0 +1 @@
+DB_PATH=../sqlite.db

pnpm-lock.yaml

@@ -42,6 +42,9 @@ importers:
       drizzle-orm:
         specifier: ^0.32.0
         version: 0.32.0(@types/better-sqlite3@7.6.9)(better-sqlite3@11.1.2)
+      ky:
+        specifier: ^1.5.0
+        version: 1.5.0
       zod:
         specifier: ^3.22.4
         version: 3.22.4
@@ -1224,6 +1227,10 @@ packages:
     resolution: {integrity: sha512-o+NO+8WrRiQEE4/7nwRJhN1HWpVmJm511pBHUxPLtp0BUISzlBplORYSmTclCnJvQq2tKu/sgl3xVpkc7ZWuQQ==}
     engines: {node: '>=6'}
 
+  ky@1.5.0:
+    resolution: {integrity: sha512-bkQo+UqryW6Zmo/DsixYZE4Z9t2mzvNMhceyIhuMuInb3knm5Q+GNGMKveydJAj+Z6piN1SwI6eR/V0G+Z0BtA==}
+    engines: {node: '>=18'}
+
   lilconfig@2.1.0:
     resolution: {integrity: sha512-utWOt/GHzuUxnLKxB6dk81RoOeoNeHgbrXiuGk4yyF5qlRz+iIVWu56E2fqGHFrXz0QNUhLB/8nKqvRH66JKGQ==}
     engines: {node: '>=10'}
@@ -2675,6 +2682,8 @@ snapshots:
 
   kleur@4.1.5: {}
 
+  ky@1.5.0: {}
+
   lilconfig@2.1.0: {}
 
   lilconfig@3.1.1: {}

rust/Cargo.lock

@@ -353,6 +353,7 @@ dependencies = [
  "iana-time-zone",
  "js-sys",
  "num-traits",
+ "serde",
  "wasm-bindgen",
  "windows-targets 0.52.6",
 ]

rust/Cargo.toml

@@ -9,7 +9,7 @@ edition = "2021"
 again = "0.1.2"
 anyhow = "1.0.79"
 base64 = "0.21.7"
-chrono = "0.4"
+chrono = { version = "0.4", features = ["serde"] }
 clap = { version = "4.4.15", features = ["derive"] }
 cron = "0.12.0"
 sqlx = { version = "0.8", features = [ "runtime-tokio", "sqlite", "chrono", "json" ] }

rust/src/api/main.rs

@@ -1,4 +1,11 @@
-use axum::{extract::State, http::StatusCode, response::IntoResponse, routing::get, Json, Router};
+use axum::{
+    extract::{Path, State},
+    http::StatusCode,
+    response::IntoResponse,
+    routing::get,
+    Json, Router,
+};
+use chrono::{DateTime, Utc};
 use clap::ValueEnum;
 use futures::future::join_all;
 use itertools::Itertools;
@@ -168,6 +175,104 @@ async fn get_best_selling(State(pool): State<SqlitePool>) -> impl IntoResponse {
     Json(categories_with_products)
 }
 
+async fn get_product_history(
+    State(pool): State<SqlitePool>,
+    Path(ean): Path<String>,
+) -> impl IntoResponse {
+    #[derive(sqlx::FromRow, Debug, Serialize)]
+    struct Precio {
+        ean: String,
+        fetched_at: chrono::DateTime<Utc>,
+        precio_centavos: Option<i64>,
+        in_stock: Option<bool>,
+        url: String,
+        name: Option<String>,
+        image_url: Option<String>,
+    }
+
+    let precios = sqlx::query!(
+        "
+select ean,fetched_at,precio_centavos,in_stock,url,name,image_url from precios
+where ean = ?
+order by fetched_at
+",
+        ean
+    )
+    .map(|r| Precio {
+        ean: r.ean,
+        url: r.url,
+        fetched_at: DateTime::from_timestamp(r.fetched_at, 0).unwrap(),
+        image_url: r.image_url,
+        name: r.name,
+        in_stock: r.in_stock.map(|x| x == 1),
+        precio_centavos: r.precio_centavos,
+    })
+    .fetch_all(&pool)
+    .await
+    .unwrap();
+
+    Json(precios)
+}
+async fn search(State(pool): State<SqlitePool>, Path(query): Path<String>) -> impl IntoResponse {
+    let sql_query = query
+        .clone()
+        .replace("\"", "\"\"")
+        .split(" ")
+        .map(|x| format!("\"{}\"", x))
+        .join(" ");
+
+    #[derive(Serialize)]
+    struct Result {
+        ean: String,
+        name: String,
+        image_url: String,
+    }
+
+    let results = sqlx::query!(
+        "with search_results as (
+            select f.ean from precios_fts f
+            where f.name match ? and f.ean != ''
+            group by f.ean
+			limit 100
+        )
+        select p.id, p.ean, p.name, p.image_url from search_results as s
+        join precios as p
+        on p.ean = s.ean
+        where p.fetched_at = (
+            SELECT MAX(fetched_at)
+            FROM precios as pf
+            WHERE pf.ean = s.ean and pf.name is not null
+        );",
+        sql_query
+    )
+    .fetch_all(&pool)
+    .await
+    .unwrap()
+    .into_iter()
+    .map(|r| Result {
+        ean: r.ean,
+        image_url: r.image_url.unwrap(),
+        name: r.name.unwrap(),
+    })
+    .collect_vec();
+
+    Json(results)
+}
+
+async fn get_info(State(pool): State<SqlitePool>) -> impl IntoResponse {
+    #[derive(Serialize)]
+    struct Info {
+        count: i64,
+    }
+
+    let count = sqlx::query!("select count(distinct ean) as count from precios")
+        .fetch_one(&pool)
+        .await
+        .unwrap()
+        .count;
+    Json(Info { count })
+}
+
 #[tokio::main]
 async fn main() {
     tracing_subscriber::fmt::init();
@@ -205,6 +310,9 @@ async fn main() {
         .route("/", get(index))
         .route("/api/healthcheck", get(healthcheck))
         .route("/api/0/best-selling-products", get(get_best_selling))
+        .route("/api/0/ean/:ean/history", get(get_product_history))
+        .route("/api/0/info", get(get_info))
+        .route("/api/0/search/:query", get(search))
         .with_state(pool);
 
     let listener = tokio::net::TcpListener::bind("0.0.0.0:8000").await.unwrap();

rust/src/scraper/auto.rs

@@ -6,6 +6,8 @@ use crate::db::Db;
 use crate::scraper::Scraper;
 use futures::Future;
 use preciazo::supermercado::Supermercado;
+use rand::seq::SliceRandom;
+use rand::thread_rng;
 use reqwest::Url;
 
 #[derive(Clone)]
@@ -48,6 +50,7 @@ impl Auto {
             if let Some(n) = self.args.n_products {
                 links.truncate(n);
             }
+            links.shuffle(&mut thread_rng());
             links
         };
         // {

sitio/.env.development

@@ -0,0 +1,2 @@
+DB_PATH=../sqlite.db
+VITE_API_HOST=http://localhost:8000

sitio/.gitignore

@@ -4,7 +4,7 @@ node_modules
 /.svelte-kit
 /package
 .env
-.env.*
+*.local
 !.env.example
 vite.config.js.timestamp-*
 vite.config.ts.timestamp-*

sitio/package.json

@@ -40,6 +40,7 @@
     "chartjs-adapter-dayjs-4": "^1.0.4",
     "dayjs": "^1.11.10",
     "drizzle-orm": "^0.32.0",
+    "ky": "^1.5.0",
     "zod": "^3.22.4"
   },
   "packageManager": "pnpm@9.5.0+sha512.140036830124618d624a2187b50d04289d5a087f326c9edfc0ccd733d76c4f52c3a313d4fc148794a2a9d81553016004e6742e8cf850670268a7387fc220c903"

sitio/src/lib/ProductPreview.svelte

@@ -1,5 +1,9 @@
 <script lang="ts" context="module">
-  export type Product = { ean: string; name: string; image_url: string | null };
+  export type Product = {
+    ean: string;
+    name: string | null;
+    image_url: string | null;
+  };
 </script>
 
 <script lang="ts">

sitio/src/lib/index.ts

@@ -1 +1,2 @@
 // place files you want to import through the `$lib` alias in this folder.
+export const API_HOST = import.meta.env.VITE_API_HOST;

sitio/src/lib/server/db.ts

@@ -1,2 +0,0 @@
-export { getDb } from "db-datos/db.js";
-export * as schema from "db-datos/schema.js";

sitio/src/routes/+layout.server.ts

@@ -1,15 +1,17 @@
-import { countDistinct } from "drizzle-orm";
 import type { PageServerLoad } from "./$types";
-import { getDb, schema } from "$lib/server/db";
+import { z } from "zod";
-const { precios } = schema;
+import ky from "ky";
+import { API_HOST } from "$lib";
+
+async function getInfo() {
+  return z
+    .object({
+      count: z.number(),
+    })
+    .parse(await ky.get(`${API_HOST}/api/0/info`).json());
+}
 
 export const load: PageServerLoad = async () => {
-  const db = await getDb();
+  const nProductos = (await getInfo()).count;
-  const nProductosR = await db
-    .select({
-      count: countDistinct(precios.ean),
-    })
-    .from(precios);
-  const nProductos = nProductosR[0].count;
   return { nProductos };
 };

sitio/src/routes/+page.server.ts

@@ -2,7 +2,9 @@ import type { PageServerLoad } from "./$types";
 import z from "zod";
 
 async function getBestSelling() {
-  const res = await fetch("http://localhost:8000/api/0/best-selling-products");
+  const res = await fetch(
+    `${import.meta.env.VITE_API_HOST}/api/0/best-selling-products`,
+  );
   const json = await res.json();
   return z
     .array(

sitio/src/routes/ean/[ean]/+page.server.ts

@@ -1,20 +1,23 @@
 import { error } from "@sveltejs/kit";
-import { eq } from "drizzle-orm";
 import type { PageServerLoad } from "./$types";
-import { getDb, schema } from "$lib/server/db";
+import { z } from "zod";
-const { precios } = schema;
+import { zPrecio, type Precio } from "./common";
+import { API_HOST } from "$lib";
+
+async function getProductHistory(ean: string) {
+  const res = await fetch(`${API_HOST}/api/0/ean/${ean}/history`);
+  const json = await res.json();
+  return z.array(zPrecio).parse(json);
+}
 
 export const load: PageServerLoad = async ({ params }) => {
-  const db = await getDb();
+  const res = await getProductHistory(params.ean);
-  const q = db
-    .select()
-    .from(precios)
-    .where(eq(precios.ean, params.ean))
-    .orderBy(precios.fetchedAt);
-  const res = await q;
   if (res.length === 0) return error(404, "Not Found");
 
-  const meta = res.findLast((p) => p.name);
+  const meta = res.findLast(
+    (p): p is Precio & { name: string; image_url: string } =>
+      !!(p.name && p.image_url),
+  );
 
   return { precios: res, meta };
 };

sitio/src/routes/ean/[ean]/+page.svelte

@@ -1,18 +1,18 @@
 <script lang="ts">
   import { Supermercado, hosts } from "db-datos/supermercado";
-  import * as schema from "db-datos/schema";
   import type { PageData } from "./$types";
   import Chart from "./Chart.svelte";
+  import type { Precio } from "./common";
 
   export let data: PageData;
 
-  let urls: Map<Supermercado, schema.Precio>;
+  let urls: Map<Supermercado, Precio>;
   $: urls = data.precios.reduce((prev, curr) => {
     const url = new URL(curr.url);
     const supermercado = hosts[url.hostname];
     prev.set(supermercado, curr);
     return prev;
-  }, new Map<Supermercado, schema.Precio>());
+  }, new Map<Supermercado, Precio>());
 
   const classBySupermercado: { [supermercado in Supermercado]: string } = {
     [Supermercado.Dia]: "bg-[#d52b1e] focus:ring-[#d52b1e]",
@@ -30,18 +30,18 @@
 
 {#if data.meta}
   <h1 class="text-3xl font-bold">{data.meta.name}</h1>
-  <img src={data.meta.imageUrl} alt={data.meta.name} class="max-h-48" />
+  <img src={data.meta.image_url} alt={data.meta.name} class="max-h-48" />
   <div class="flex gap-2">
-    {#each urls as [supermercado, { url, precioCentavos }]}
+    {#each urls as [supermercado, { url, precio_centavos }]}
       <a
         href={url}
         rel="noreferrer noopener"
         target="_blank"
         class={`focus:shadow-outline inline-flex flex-col items-center justify-center rounded-md ${classBySupermercado[supermercado]} px-4 py-2 font-medium tracking-wide text-white transition-colors duration-200 hover:bg-opacity-80 focus:outline-none focus:ring-2 focus:ring-offset-2`}
       >
-        {#if precioCentavos}
+        {#if precio_centavos}
           <span class="text-lg font-bold"
-            >{formatter.format(precioCentavos / 100)}</span
+            >{formatter.format(precio_centavos / 100)}</span
           >
         {/if}
         <span class="text-sm">{supermercado}</span>

sitio/src/routes/ean/[ean]/Chart.svelte

@@ -1,8 +1,8 @@
 <script lang="ts">
-  import type { Precio } from "db-datos/schema";
   // import dayjs from "dayjs";
   import ChartJs from "./ChartJs.svelte";
   import { hosts, colorBySupermercado } from "db-datos/supermercado";
+  import type { Precio } from "./common";
 
   export let precios: Precio[];
 
@@ -15,15 +15,15 @@
       const ps = precios
         .filter((p) => new URL(p.url!).hostname === host)
         .filter(
-          (p): p is Precio & { precioCentavos: number } =>
+          (p): p is Precio & { precio_centavos: number } =>
-            p.precioCentavos !== null,
+            p.precio_centavos !== null,
         );
       return {
         label: supermercado,
         data: [
           ...ps.map((p) => ({
-            x: p.fetchedAt,
+            x: p.fetched_at,
-            y: p.precioCentavos / 100,
+            y: p.precio_centavos / 100,
           })),
           // lie
           // ...ps.map((p) => ({

sitio/src/routes/ean/[ean]/common.ts

@@ -0,0 +1,12 @@
+import { z } from "zod";
+
+export const zPrecio = z.object({
+  ean: z.string(),
+  fetched_at: z.coerce.date(),
+  precio_centavos: z.number().nullable(),
+  in_stock: z.boolean().nullable(),
+  url: z.string(),
+  name: z.string().nullable(),
+  image_url: z.string().nullable(),
+});
+export type Precio = z.infer<typeof zPrecio>;

sitio/src/routes/search/+page.server.ts

@@ -1,26 +1,29 @@
-import { sql } from "drizzle-orm";
+import { z } from "zod";
 import type { PageServerLoad } from "./$types";
-import { getDb } from "$lib/server/db";
+import { API_HOST } from "$lib";
+import ky from "ky";
+
+const zProductResult = z.object({
+  ean: z.string(),
+  name: z.string(),
+  image_url: z.string(),
+});
+
+async function search(query: string) {
+  return z
+    .array(zProductResult)
+    .parse(
+      await ky
+        .get(`${API_HOST}/api/0/search/${encodeURIComponent(query)}`)
+        .json(),
+    );
+}
 
 export const load: PageServerLoad = async ({ url }) => {
-  const db = await getDb();
   const query = url.searchParams.get("q");
-  let results: null | { ean: string; name: string; imageUrl: string }[] = null;
+  let results: null | { ean: string; name: string; image_url: string }[] = query
-  if (query) {
+    ? await search(query)
-    const sQuery = query
+    : null;
-      .replaceAll(`"`, `""`)
-      .split(" ")
-      .map((s) => `"${s}"`)
-      .join(" ");
-    console.debug(sQuery);
-    const sqlQuery = sql`select p.ean, p.name, p.image_url as imageUrl from precios_fts f
-      join precios p on p.ean = f.ean
-      where f.name match ${sQuery}
-      group by p.ean
-      having max(p.fetched_at)
-      order by p.in_stock desc;`;
-    results = db.all(sqlQuery);
-  }
 
   return { query, results };
 };